← Models

Grok Code Fast 1

xAI · Grok

Deprecated xAI coding model whose old slug now maps toward Grok Build 0.1 for coding-agent workloads.

Part of Grok family · Other versions: Grok 4.5 , Grok Build 0.1 , Grok 4.3 , Grok 4.20 , Grok 4.1 Fast , Grok 4
Type
language
Context
256K tokens
Max Output
256K tokens
Status
deprecated
Input
$1/1M tok
Output
$2/1M tok
API Access
Yes
License
proprietary
coding agentic tool-use software-engineering reasoning fast deprecated
Released August 2025 · Updated June 8, 2026

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last verified on June 8, 2026.

Grok Code Fast 1 is now a deprecated xAI coding-specialized model. xAI’s May 2026 model-retirement guide says grok-code-fast-1 retired from the API on May 15, 2026 at 12:00 PM PT, and its code-workload recommendation is to migrate to Grok Build 0.1. xAI’s current Grok Build 0.1 model page also lists grok-code-fast-1, grok-code-fast, and grok-code-fast-1-0825 as aliases.

This page remains useful for teams auditing older editor or agent integrations, but it should not be used as a new-work recommendation.

Capabilities

xAI’s launch materials emphasized the model’s fit for everyday agentic coding:

  • Repository-aware coding workflows with common tools like grep, terminal commands, and file editing.
  • Fast implementation loops for bug fixes, refactors, feature slices, and codebase questions.
  • Strong support for mainstream languages including TypeScript, Python, Java, Rust, C++, and Go.
  • Good instruction following for scoped engineering tasks where speed matters as much as quality.
  • Practical fit inside agentic coding surfaces such as Cursor, GitHub Copilot, Cline, Windsurf, and similar tools.

That historical positioning is now superseded by the Grok Build route. For concrete and iterative coding work, route new xAI callers to grok-build-0.1 explicitly rather than depending on an old slug or retirement-era redirect behavior.

Technical Details

xAI described Grok Code Fast 1 as a coding-first reasoning model designed around native tool calling. Current xAI docs instead expose Grok Build 0.1 as the coding route and alias the old Grok Code Fast names there. Historical anchors:

  • 256K context window.
  • Text-only input in the public pricing card.
  • A coding-oriented reasoning profile tuned for agentic editor workflows.

Like other older xAI model pages in this repo, maxOutput is recorded as the documented context ceiling for quick comparison. Teams should still set explicit output limits in their tooling rather than relying on that number as a guaranteed generation cap.

Pricing & Access

Current replacement-route pricing for Grok Build 0.1 is:

  • Input: $1.00 per 1M tokens
  • Cached input: $0.20 per 1M tokens
  • Output: $2.00 per 1M tokens

Historical access options:

  • xAI API
  • Launch-partner editor and coding-agent platforms
  • xAI ecosystem tooling where Grok coding routes are exposed

xAI’s launch materials also made the model temporarily free on selected partner platforms, but teams should budget based on the published API pricing instead of promotional rollout terms.

Best Use Cases

Use this page for:

  • identifying old grok-code-fast-1 calls
  • comparing historical behavior with Grok Build 0.1
  • updating coding-agent configs and docs
  • removing assumptions tied to the old coding-specific price point

Do not start new coding deployments on Grok Code Fast 1. xAI’s code-workload replacement guidance is Grok Build 0.1, while cross-provider coding tools such as OpenAI Codex and Claude Code remain separate product choices.

Comparisons

  • Grok Build 0.1 (xAI): Recommended xAI replacement for coding and web-development workloads.
  • Grok 4.3 (xAI): Better default for general reasoning and non-coding assistant workloads.
  • GPT-5.3-Codex (OpenAI): Coding-specialized OpenAI route with first-party Codex product support.
  • Claude Code (Anthropic): Tool surface rather than a single model page, but often the more practical comparison for terminal-based agentic coding.